Statements (52)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:model
|
| gptkbp:abbreviation |
gptkb:AGA
|
| gptkbp:abilities |
game playing
language understanding multimodal reasoning vision tasks robotics control |
| gptkbp:announced |
2024-05-14
|
| gptkbp:architecture |
Transformer-based
|
| gptkbp:demonstratedTask |
gptkb:navigation
language translation object recognition video understanding multi-step reasoning image captioning visual question answering instruction following Atari game playing real-world environments robotic arm manipulation robotic pick-and-place simulated environments |
| gptkbp:demonstrates |
gptkb:Google_I/O_2024
|
| gptkbp:developedBy |
gptkb:Google_DeepMind
|
| gptkbp:input |
gptkb:DVD
gptkb:illustrator gptkb:text robotic sensor data |
| gptkbp:language |
English
multilingual support |
| gptkbp:notableFeature |
scalable architecture
end-to-end learning generalization across tasks real-world robotics integration single model for multiple domains |
| gptkbp:output |
gptkb:text
actions predictions robotic control signals |
| gptkbp:purpose |
to perform a wide range of tasks across different domains
|
| gptkbp:relatedTo |
gptkb:Gemini
gptkb:RT-2 gptkb:Gato |
| gptkbp:researchPaper |
https://arxiv.org/abs/2405.09882
|
| gptkbp:trainer |
multimodal datasets
game environments language corpora robotics data |
| gptkbp:website |
https://deepmind.google/technologies/aga/
|
| gptkbp:bfsParent |
gptkb:NeurIPS_2022
|
| gptkbp:bfsLayer |
7
|
| https://www.w3.org/2000/01/rdf-schema#label |
A Generalist Agent
|